Determining reverberation time

Abstract

Included in the embodiments of the present invention is a technique to detect sound with a sensor to generate a corresponding sound signal and iteratively determine two or more values with a maximum likelihood function for evaluation of reverberation time. One of these values corresponds to a time constant parameter, and another of these values corresponds to a diffusive power parameter. An estimate representative of the reverberation time is further provided as a function of an order-statistics filter.

What is claimed is:  
     
         1 . A method, comprising: 
 detecting sound with a sensor to generate a corresponding sensor signal;    generating data with the sensor signal in accordance with a maximum likelihood estimator; and    filtering the data with an order-statistics filter to provide an estimate of reverberation time.    
     
     
         2 . The method of  claim 1 , which includes iteratively determining a decay time parameter and a power parameter during execution of said generating.  
     
     
         3 . The method of  claim 1 , which includes providing the reverberation time to one or more of a hearing assistance data processing routine, a voice recognition data processing routine, a hands-free telephony data processing routine, a teleconference data processing routine, and a sound level evaluation data processing routine.  
     
     
         4 . The method of  claim 1 , wherein said generating includes calculating a number of reverberation time parameter estimations with the maximum likelihood estimator, the estimations each being calculated as a function of a sequence of sound observations over a different time window.  
     
     
         5 . A method, comprising: 
 detecting sound with a sensor to generate a corresponding sound signal;    iteratively determining two or more values with a maximum likelihood function to evaluate one or more reverberation characteristics of an acoustic environment, one of the values corresponding to a time-constant parameter and another of the values corresponding to a diffusive power parameter; and    providing an estimate corresponding to reverberation time of the acoustic environment.    
     
     
         6 . The method of  claim 5 , wherein said iteratively determining is performed for each of a number of different frequency ranges of the sound and includes calculating a reverberation time estimate for each of the different frequency ranges.  
     
     
         7 . The method of  claim 5 , which includes providing the estimate of reverberation time to one or more of a hearing assistance data processing routine, a voice recognition data processing routine, a hands-free telephony data processing routine, a teleconference data processing routine, and a sound level evaluation data processing routine.  
     
     
         8 . The method of  claim 5 , wherein said iteratively determining is performed with each of a number of different data sequences to provide a number of reverberation time estimations, the data sequences each being representative of a series of sound observations over a different time window.  
     
     
         9 . The method of  claim 8 , wherein said providing includes filtering the reverberation time estimations with an order-statistics filter to select the estimate corresponding to reverberation of the acoustic environment.  
     
     
         10 . A method, comprising: 
 preparing a number of observed sound data sequences each representative of sound over one of a number of different time periods;    determining each of a number of estimations as function of a different one of the sequences in accordance with a maximum likelihood estimator; and    selecting one of the estimations to provide a reverberation time.    
     
     
         11 . The method of  claim 10 , wherein said selecting includes filtering the estimations.  
     
     
         12 . The method of  claim 11 , wherein said filtering is performed with an order-statistics filter.  
     
     
         13 . The method of  claim 10 , wherein said determining includes iteratively calculating each of at least two values, a first one of the values corresponding to a decay time and a second one of the values corresponding to diffusive power.  
     
     
         14 . The method of  claim 13 , wherein said iteratively calculating is performed for each of a number of different frequency ranges of the sound and includes calculating a reverberation time estimate for each of the different frequency ranges.  
     
     
         15 . The method of  claim 10 , which includes providing the one of the estimations to one or more of a hearing assistance data processing routine, a voice recognition data processing routine, a hands-free telephony data processing routine, a teleconference data processing routine, and a sound level evaluation data processing routine.  
     
     
         16 . The method of  claim 10 , wherein the sequences each correspond to a different series of observed sound samples and the different time periods each correspond to a different time window.  
     
     
         17 . The method of  claim 16 , which includes adaptively changing the duration of the different time windows.  
     
     
         18 . A method, comprising: 
 preparing a number of observed sound data sequences each corresponding to sound over a different one of a number of time windows;    determining a number of reverberation time parameter estimations, the estimations each being determined as a function of a different one of the sequences in accordance with a parameter estimator; and    filtering the estimations with an order-statistics filter.    
     
     
         19 . The method of  claim 18 , wherein the parameter estimator is based on maximum likelihood estimation.  
     
     
         20 . The method of  claim 18 , wherein the parameter estimator is of a robust type.  
     
     
         21 . The method of  claim 18 , wherein said determining includes iteratively calculating each of at least two of the parameters for each of the sequences, a first one of the parameters corresponding to a decay time and a second one of the parameters corresponding to power.  
     
     
         22 . The method of  claim 21 , wherein said iteratively calculating is performed for each of a number of different frequency ranges of the sound and includes calculating a reverberation time parameter estimate for each of the different frequency ranges.  
     
     
         23 . The method of  claim 18 , which includes selecting one the estimations based on said filtering.  
     
     
         24 . The method of  claim 23 , which includes providing the one of the estimations to one or more of a hearing assistance data processing routine, a voice recognition data processing routine, a hands-free telephony data processing routine, a teleconference data processing routine, and a sound level evaluation data processing routine.  
     
     
         25 . The method of  claim 18 , which includes adaptively changing the duration of the time windows.  
     
     
         26 . A system, comprising: 
 a sensor for detecting sound; and    a processing subsystem operable to receive sound-representative signals from the sensor to determine a reverberation time estimate of an unknown acoustic environment by processing the signals with a parameter estimator and order-statistics filter.    
     
     
         27 . The system of  claim 26 , wherein the parameter estimator is of robust type.  
     
     
         28 . The system of  claim 26 , wherein the parameter estimator is of a maximum likelihood type.  
     
     
         29 . The system of  claim 26 , further comprising means for processing the reverberation time estimate in one or more of a hearing assistance data processing routine, a voice recognition data processing routine, a hands-free telephony data processing routine, a teleconference data processing routine, and a sound level evaluation data processing routine.  
     
     
         30 . The system of  claim 26 , wherein the subsystem includes: 
 means for processing a number of sequences of observed sound data to provide a corresponding number of reverberation time parameter estimations with the parameter estimator, the sequences each corresponding to one of a number of different time windows; and    means for filtering the estimations with the order-statistics filter to provide the reverberation time estimate.    
     
     
         31 . The system of  claim 30 , further comprising at least one of: 
 means for performing the processing means for each of a number of different frequency ranges of the sound; and    means for adaptively changing the duration of the time windows.    
     
     
         32 . An apparatus, comprising: a device carrying logic executable by a processor to process data corresponding to a number of sequences of sound samples, the sequences each being representative of sound over a different time period, the logic being further operable to determine a number of reverberation time parameter estimations each as function of a different one of the sequences in accordance with a maximum likelihood estimator and filter the estimations to provide a selected reverberation time estimate.  
     
     
         33 . The apparatus of  claim 32 , wherein the logic includes a number of software instructions and the device includes a computer-readable memory storing the software instructions.  
     
     
         34 . The apparatus of  claim 32 , wherein the device includes one or more parts of a computer network and the logic is encoded in one or more signals by the device.  
     
     
         35 . The apparatus of  claim 32 , wherein the logic is operable to filter the estimations with an order-statistics type of filter.